| Reader discretion is advised :) The advice given by the current article is just after the "followed by occasional crashes" state of the state graph : The author has already noticed that the "edge" is context dependent and still recommends winging it (he also mention previous personal problems with substance abuse so we are definitely in the "here be dragons" territory). The author seems to be in the wreck everyone around you and then repair it business. ** There are various culture on the internet, notably pushed by influencers and other self proclaimed gurus, that often push it further that reasonable, often amplified by filter bubbles and echo chamber that algorithm like to recommend. It's typically called Bro-Science in the fitness domain. This empirical science has the benefit of the wisdom of the crowd. Typically in running or fitness, one follow a training program to make some gains. But the danger is often over-training, or bad posture resulting in an injury. Learning unguided is often the cause of injury. Typically what happens is your mental model you are basing your training program on is missing some important levers : For example you didn't consider some cumulative nutrient deficiency or adaptation that occur when not respecting cycles by having rest weeks. But you push yourself because you don't see a reason why you shouldn't push yourself and then the injury occur. ** Real science here would recommend taking a step back, and look for the missing lever instead of random pushing the levers in front of you. You should be humble and start with the hypothesis that there is some additional unknown risk you aren't aware of. Reasonable progress isn't made working on the frontier trying to push it forward, but rather by exploring more diversity while staying comfortably in the comfort zone in order to be able to handle the occasional rough patches that natural variance will throw at you. The science behind if you want to learn more is https://en.wikipedia.org/wiki/Exploration-exploitation_dilem... and https://en.wikipedia.org/wiki/Multi-armed_bandit (Thomson sampling and no regret sampling). It typically consist of building an inner representation of the world (aka a world model) and acting on it conservatively. You may also look into martingale stopping time if you are inclined in probability theory. |